Ganglioside GD3 Traffics from the trans- Golgi Network to Plasma Membrane by a Rab11-independent and Brefeldin A-insensitive Exocytic Pathway*

Gangliosides, complex glycosphingolipids containing sialic acids, have been found to reside in glycosphingo-lipid-enriched microdomains (GEM) at the plasma membrane. They are synthesized in the lumen of the Golgi complex and appear unable to translocate from the lumenal toward the cytosolic surface of Golgi membrane to access the monomeric lipid transport. As a consequence, they can only leave the Golgi complex via the lumenal surface of transport vesicles. In this work we analyzed the exocytic transport of the disialo ganglioside GD3 from trans -Golgi network (TGN) to plasma membrane in CHO-K1 cells by immunodetection of endogenously synthesized GD3. We found that ganglioside GD3, unlike another luminal membrane-bounded lipid (glycosylphos-phatidylinositol-anchored protein), did not partition into GEM domains in the Golgi complex and trafficked from TGN to plasma membrane by a brefeldin A-insensitive exocytic pathway. Moreover, a dominant negative form of Rab11, which prevents exit of vesicular stomatitis virus glycoprotein of the exit GPI-YFP complex, (37 (cid:1) BFA (37 °C, (cid:4) BFA of 2 (cid:4) g/ml BFA for 1 h. BFA in the culture min of , Sial-T2-expressing CHO-K1 incubated for h inabsence ( (cid:1) BFA or ( (cid:4) BFA ) of (cid:4) g/ml immunostained anti-HA by the appropri- ate second intrinsic fluorescence of YFP the

ing distally in subapical compartments in HepG2 cells (1) and to regulate the delivery of proteins from the trans-Golgi network (TGN) to the plasma membrane in both polarized cells and fibroblasts (11)(12)(13)(14)(15). On the other hand the endocytic transport of glycosphingolipids to the Golgi complex is dependent upon Rab7 and Rab9, as demonstrated by inhibition of Golgi targeting in normal cells by dominant-negative Rab7 and Rab9 constructs (16).
In addition to the Rab GTPases, ARF GTPases are required for intracellular transport and maintaining the integrity of organelle structure. Mammalian ARF proteins are grouped into three classes; class I includes ARF1, -2, and -3, class II includes ARF4 and -5, and class III includes ARF6. ARF1 has been the most extensively characterized and is established to be involved in the budding of vesicles from the Golgi complex. On the other hand ARF6 has been implicated in the formation of vesicles from plasma membranes (17). ARF activation and inactivation are catalyzed by guanine nucleotide exchange factors (GEFs) that facilitate GTP binding and GTPase-activating proteins that catalyze GTP hydrolysis. In general ARF6 GEFs are not inhibited by brefeldin A (BFA), in contrast with other ARF GEFs (18). BFA is a metabolite of the fungus Eupenicillium brefeldianum that acts by a rare uncompetitive mechanism with formation of an abortive ARF⅐GDP-BFA-GEF complex, which cannot proceed to nucleotide dissociation (18).
Glycolipid transport, particularly ganglioside transport, is very poorly studied. Short chain and analogous fluorescent lipids have been used as useful tools for delineating potential pathways and mechanisms of intracellular transport and sorting (19). However, the quantitative and qualitative behavior of the analog lipids is quite different from long chain cellular lipids, since the ability of short fluorescent lipids to diffuse spontaneously between different membranes is not generally shared by their endogenous lipid counterparts (20).
In this study we analyzed the transport of the disialo ganglioside GD3 from TGN to plasma membrane in CHO-K1 cells by immunodetection of endogenously synthesized GD3. Our main approach was to accumulate GD3 at the TGN with temperature block and allow the synchronous release upon warming. GD3 arrival at cell surface was easily monitored by immunodetection since gangliosides from plasma membranes were previously depleted by incubating the cells with a potent inhibitor of complex glycolipid synthesis. We found that GD3 ganglioside, unlike another luminal membrane-bounded lipid (glycosylphosphatidylinositol-anchored protein) (21)(22)(23)(24), did not partition into GEM domains in the Golgi complex and trafficked from TGN to plasma membrane via a BFA-insensitive exocytic pathway. Moreover, the dominant negative form of Rab11, which prevents the exit of vesicular stomatitis virus glycoprotein (VSVG) from the Golgi complex, did not influence the capacity of GD3 to reach the cell surface.
Taken together our results strongly support the notion that most GD3 ganglioside traffics from TGN to plasma membrane by a non-conventional vesicular pathway where lateral membrane segregation of VSVG (non-GEM resident) and GPI-anchored proteins (GEM resident) from GD3 is required before exiting TGN.
Cell Culture and DNA Transfections-CHO-K1 cells (clone 2), a stable CMP-NeuAc:GM3 sialyltransferase (Sial-T2, tagged at the C terminus with the nanopeptide epitope of the viral hemagglutinin (HA)) transfectant expressing the ganglioside GD3 and GT3 (26,27) were used. Cells were grown and maintained at 37°C in 5% CO 2 in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum and antibiotics. Where indicated cells were transfected with 1 g/ 35-mm dish of the plasmids named above using cationic lyposomes (LipofectAMINE, Invitrogen) according to the manufacturer's instructions and incubated for 14, 24, or 48 h at 37°C with the transfection reagent and DNA mixture. When VSVG-YFP plasmid was used for transfections, CHO-K1 cells grown in the presence of the inhibitor of glycolipid synthesis were transfected at 41°C (restrictive temperature) for 14 h. Under these conditions the VSVG protein accumulated in the endoplasmic reticulum (ER). Cells transfected with VSVG-YFP or GPI-YFP were then incubated at 20°C for 2 h and either processed for immunofluorescence immediately or transferred to 37°C for 1 or 2 additional hours before fixation. Where indicated, cells were supplemented with BFA from a stock solution 1 mg/ml in methanol at a final concentration of 2 g/ml. BFA was present in the culture medium for the last 15 min of culture at 20°C and for 1 h at 37°C. Control cells were supplemented with the same amount of methanol without BFA. Cycloheximide (30 g/ml) was added to the culture medium for the last 30 min of culture at 20°C and maintained for the indicated times.

D,L-threo-1-Phenyl-2-hexadecanoylamino-3-pyrrolidino-1-propanol-
HCl Treatment of Cells-Inhibition of glycolipid synthesis with D,Lthreo-1-phenyl-2-hexadecanoylamino-3-pyrrolidino-1-propanol-HCl (P4) (Matreya Inc., Pleasant Gap, PA) was carried out as follows. Cells from clone 2 in culture were treated for 5 days with 2.4 M P4 added to the culture medium from a recently prepared 600 M stock solution. After 3 days of treatment with P4, cells were further incubated in Dulbecco's modified Eagle medium without serum growth factors for 34 h to synchronize them. Then, to stimulate cell growth, cells were incubated for 14 h with fresh medium containing 10% fetal bovine serum. After 5 days of treatment with P4, culture medium was removed, and cells were washed extensively with PBS to remove P4. Then, cells were incubated with 10% fetal bovine serum-rich medium for 2 h at 20°C to accumulate GD3 in TGN and immediately incubated at 37°C to allow GD3 transport.
Confocal Immunofluorescence Microscopy-Cells grown in coverslips were washed twice in PBS, fixed in acetone at Ϫ20°C for 7 min, washed in PBS, and incubated in 3% bovine serum albumin, PBS buffer for 1 h at 37°C to block nonspecific binding sites. Coverslips were then incubated overnight at 4°C with primary antibodies, washed 5 times with 1% bovine serum albumin, PBS buffer, and exposed to secondary antibodies for 90 min at 37°C. The primary antibodies were: mouse monoclonal antibody anti-GD3 (IgG3), clone R24 (a gift of Dr. Confocal images were collected using a Carl Zeiss LSM5 Pascal laser-scanning confocal microscope equipped with an argon/helium/ neon laser and a 63ϫ (numerical aperture ϭ 1.4) oil immersion objective (Zeiss Plan-Apochromat). Single confocal sections of 0.7 m were taken parallel to the coverslip (xy sections). Images were acquired and processed with the Zeiss LSM image software. Images for fluorescence quantification were taken with a Zeiss inverted microscopy equipped with epifluorescence and processed with the MetaMorph 3.0 Imagen System (Universal Imaging, West Chester, PA). Final images were compiled with Adobe Photoshop 5.0.
Lipid Labeling, Extraction, and Chromatography-For metabolic labeling ex vivo, cells in culture (3 ϫ 10 5 cells/35-mm dish) were labeled with 10 Ci/ml [9,10(n)-3 H]palmitic acid (Amersham Biosciences; 53 Ci/mmol) during 12 h in the presence of the inhibitor P4 to block synthesis of GD3. Then, culture medium was removed; cells were washed extensively to remove P4 and incubated with complete medium for 2 h at 20°C to synthesize and accumulate GD3 in TGN. Following 20°C treatment, cells were incubated at 37°C to allow GD3 exit from TGN and the detergent solubility/insolubility of GD3 was analyzed at different times. Cells were washed with cold PBS and harvested by scraping. Samples were treated with 0.5 ml lysis buffer containing 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.1 M Na 2 CO 3 , 5 g/ml aprotinin, 0.5 g/ml leupeptin, 0.7 g/ml pepstatin and 25 mM Tris-HCl pH 7.5 (TNE buffer) at 4°C for 1h, and then centrifuged for 1h at 100000 g at 4°C. The supernatant (soluble fraction) was removed, and the pellet (insoluble fraction) was resuspended in 0.2 ml lysis buffer. Glycolipids from the supernatant (soluble fraction) and from the pellet (insoluble fraction) were subjected to Folch partition (28). The resulting aqueous phases were freed from Triton X-100 (non-ionic molecule) by passing through DEAE-Sephadex column. Under this condition, neutral lipids as glucosylceramide and LacCer co-eluted with Triton X-100 in the non-retained fraction. The acidic lipid fraction retained in the column (gangliosides) was eluted with 0,1 M sodium acetate and the salt removed by passing through Sep-Pak C 18 cartridge column (Waters Corporation Milford, USA). The eluted lipid fraction was supplemented with appropriate amounts of standard gangliosides and chromatographed on high-performance TLC plates (HPTLC, Merck, Germany) using chloroform, methanol, 0.2% CaCl 2 (60:36:8 v/v) as solvent. Standard gangliosides were visualized by exposure of the plate to iodine vapors. Routinely 700 -1000 cpm were spotted on each lane. Radioactive gangliosides were visualized using a Fuji Photo Film Bio Imagen analyzer or visualized by fluorography after dipping the plate in 0.4% melted 2,5-diphenyloxazole in 2-methylnaphthalene and exposure to a radiographic film at Ϫ70°C, usually for 15 days.
For analysis of sphingomyelin and ceramide, cells in culture were labeled with 10Ci/ml of [9,10(n)-3 H]palmitic acid for 18 h in the absence or presence of the inhibitor P4 as described (29,30). After incubation, the radioactive medium was removed, the cells were washed, and the lipids were extracted by the method of Folch et al. (28). The resulting aqueous phase (upper phase) was freed from water-soluble contaminants by passing through a Sephadex G-25 column. Lipids from the upper phase were spotted on a HPTLC plate and developed in chloroform, methanol, 0.2% CaCl 2 (60:36:8 v/v). The chloroform phase (lower phase) was subjected to mild alkaline methanolysis to remove glycerophospholipids, and lipids were extracted once more by the method of Folch et al. (28). The chloroform phase was dried under nitrogen, and lipids were separated by HPTLC. For HPTLC analysis of ceramide, samples were spotted on an HPTLC plate and developed in chloroform, acetic acid (9:1 v/v). Ceramide (type III; Sigma) was used as a standard. For HPTLC analysis of sphingomyelin, samples were spotted on an HPTLC plate and developed in chloroform, acetone, acetic acid, methanol, water (50:20:15:10:5 v/v). Bovine sphingomyelin was used as standard. Standards lipids were visualized by exposure of the plate to iodine vapors. Radioactive lipids were visualized by fluorography.

RESULTS
GD3 Cell Depletion, de Novo Resynthesis, and TGN Accumulation at 20°C-To investigate GD3 transport from TGN to plasma membrane, we developed a successful model in GD3expressing CHO-K1 cell clone, already established in our laboratory (23,26). We could analyze the transport to plasma membrane of natural GD3, synthesized de novo, in cells where more than 96% of total GD3, mainly present in plasma membrane, was depleted by using P4, a potent inhibitor of complex glycolipid synthesis (23, 29) (see the scheme in Fig. 2). Cells were grown in the presence of P4 for 5 days, allowing the cells to eliminate GD3 from plasma membrane. Under this condition, the analysis of GD3 expression by immunostaining with monoclonal antibody R24 was under the limit of detection (Fig. 1A). After 5 days of treatment with P4, culture medium was removed; cells were washed extensively to remove P4 and incubated with complete medium for 2 h at 20°C, which causes proteins to accumulate in the TGN (31). At the end of this period, cells were able to synthesize and accumulate GD3 in TGN (Fig. 1B), since GD3 was found co-localizing to a significant extent with UDP-GalNAc: LacCer/GM3/GD3 N-acetylgalactosaminyltransferase and mannose 6-phosphate receptor, two well characterized TGN markers in CHO-K1 cells (26, 32) (data not shown). To our knowledge this is the first time in which a natural ganglioside synthesized de novo is accumulated and immunodetected in TGN. After a 20°C treatment, cells were incubated at 37°C for 7, 15, 30, and 90 min to allow GD3 exit and transport from TGN to the plasma membrane ( Fig. 1, C-F). GD3-expressing cells showed this ganglioside at the cell surface as patches, although a minor fraction was also present in late endosomes (24). As a further evaluation of the subcellular distribution of GD3 at different temperatures and times, quantification of GD3 fluorescence intensity was performed (Table I). The results obtained confirmed our qualitative observation and clearly revealed that the pool of GD3 localized in Golgi complex (at 20°C) was reduced progressively at 37°C, whereas a noticeable accumulation of the glycolipid was observed in the plasma membrane. Next, we set out to confirm the decrease of GD3 after P4 treatment and its effect on endogenous ceramide and sphingomyelin levels, studying the content of these lipids after metabolic labeling with [ 3 H]palmitic acid. Biochemical results from Fig. 2 clearly shows that exposure of cells to P4 in the culture medium for 5 days led to a noticeable decrease of both GD3 and GT3 ( Fig. 2A) and their precursors lactosylceramide (LacCer) and glucosylceramide (GlcCer) (Fig. 2B). On the FIG. 1. Accumulation of GD3 in TGN at 20°C and latter transport to the cell surface of CHO-K1 cells. Cells were grown in the presence of P4 for 5 days, allowing the cells to mainly eliminate GD3 from plasma membrane (A). Then culture medium was removed, and cells were washed extensively to remove P4 and incubated with complete medium for 2 h at 20°C to accumulate GD3 in the TGN (B). At the end of this period cells were incubated at 37°C for during 7 (C), 15 (D), 30 (E), and 90 (F) min to allow GD3 exit and transport from TGN to plasma membrane. GD3 expression was analyzed by confocal microscopy immunofluorescence in acetone-fixed cells using a monoclonal antibody R24 as the primary antibody and Alexa 546-conjugated goat anti-mouse IgG as the secondary antibody. other hand the presence of the inhibitor in the culture medium of CHO-K1 cells did not significantly alter the endogenous levels of sphingomyelin (Fig. 2B) and ceramide (Fig. 2C). Thus, we have established the experimental conditions to explore the role of different molecules of the sorting machinery in GD3 traffic to plasma membrane.
The Disialo Ganglioside GD3 Traffics to Plasma Membrane by a Rab11-independent Route-Rab11 is a Ras-related small GTPase that has been shown to regulate the delivery of proteins from the TGN to the plasma membrane either through endosomes or by a direct transit (13). Before evaluating the role of Rab11 on GD3 transport from TGN to plasma membrane, we investigated in CHO-K1 cells the effect of these GTPases on the sorting to plasma membrane of the temperature-sensitive VSVG protein, a useful transport marker. GD3-expressing CHO-K1 cells were grown in the presence of P4 for 5 days, and VSVG was expressed by transient DNA transfection 14 h before removing the inhibitor at 41°C to accumulate VSVG in the ER. Then cells were incubated at 20°C for 2 h to allow the accumulation of VSVG in the TGN in media containing cycloheximide to prevent further synthesis of VSVG (Fig. 3A, 20°C). VSVG transport to the cell surface was initiated by changing the temperature to a permissive condition (Fig. 3A, 37°C). Co-expression of the dominant negative Rab11S25N protein (Rab11S25N-GFP) was found to prevent the exit of VSVG protein from the Golgi complex, whereas wild type Rab11 (Rab11wt-GFP) protein expression allowed normal delivery of the protein marker to the plasma membrane (Fig. 3A, 37°C). The expression of the constitutively activated Rab11 protein (Rab11Q70L) had no appreciable effect on VSVG delivery to the plasma membrane (data not shown). As also shown in Fig. 3, Rab11wt and Rab11S25N proteins were expressed both in the perinuclear region of the cells and dispersed throughout the cytoplasm. The perinuclear expression of Rab11wt-GFP was observed to be mainly associated with recycling endosomes, whereas Rab11S25N-GFP was found co-localizing with Golgi complex and recycling endosome markers. 2 The expression of Rab11 variants did not alter the Golgi complex morphology, as evidenced by for the expression of Sial-T2, a Golgi proximal resident (Fig. 3C). The analysis of Rab11 protein expression by Western blotting revealed the expected molecular mass (55 kDa) (data not shown).
Having shown that Rab11 participates in the delivery of a protein cargo (VSVG) from Golgi complex to plasma membrane in CHO-K1 cells, we next asked whether this protein might be involved in the transport of the disialo ganglioside GD3 between these organelles. To explore this we basically applied the protocol described in Fig. 1, except that Rab proteins were expressed by transient DNA transfection 2 days before removing the inhibitor of ganglioside synthesis (P4). After 2 h at 20°C in the absence of the P4 inhibitor, GD3 synthesized de novo was mainly accumulated at the TGN (Fig. 3B, 20°C). Under these conditions, the expression of wild type and mutant forms of Rab11 proteins had no apparent effect on cell physiology and GD3 accumulation in Golgi complex. Transfer to 37°C caused the GD3 to become progressively concentrated at the plasma membrane in non-transfected cells (Fig. 3B, 37°C), as shown in Fig. 1. Interestingly, the expression of wild type and dominant negative forms of Rab11 did not influence the capacity of GD3 to reach the cell surface (Fig. 3B). Moreover, the expression of the constitutively activated Rab11 protein (Rab11Q70L) had no appreciable effect on GD3 delivery to the plasma membrane (data not shown).  latter transport to cell surface at 37°C Cells were grown in the presence of P4 for 5 days, allowing the cells to mainly eliminate GD3 from plasma membrane. Then culture medium was removed; cells were washed extensively to remove P4 and incubated for 2 h at 20°C to accumulate GD3 in the TGN. At the end of this period cells were incubated at 37°C for 7, 15, 30, 90, and 120 min to allow GD3 exit and transport from TGN to plasma membrane. GD3 expression was analyzed by immunofluorescence in acetone-fixed cells using a monoclonal antibody R24 as the primary antibody and Alexa 546-conjugated goat anti-mouse IgG as the secondary antibody. The cells were analyzed with a Zeiss inverted microscopy equipped with epifluorescence. Each value is the mean Ϯ S.D. At least 40 cells were measured for each experimental condition. Golgi membrane is sensitive to BFA (33) and is most likely regulated by the TGN-localized and BFA-inhibited GEFs (34). Before testing the potential role of this machinery in the exit of GD3 from Golgi complex, we first demonstrated that the constitutive transport of VSVG in GD3-expressing CHO-K1 cells is rapidly inhibited upon treatment with BFA, as already reported for BHK-21 cells (35 (Fig. 4, A and B). Then cells were shifted to 37°C in the absence (ϪBFA) or presence (ϩBFA) of BFA for 1 h (Fig. 4, C and D). VSVG was detected by analysis of the intrinsic fluorescence of YFP fused to VSVG. As shown, the extent of VSVG that arrived FIG. 3. GD3 traffics to the plasma membrane by a Rab11-independent pathway. A, GD3-expressing CHO-K1 cells were grown in the presence of P4 for 5 days and co-transfected before removing the inhibitor P4 with plasmids carrying cDNAs coding for VSVG and wild type Rab11 (Rab11wt-GFP) proteins or for VSVG and dominant negative Rab11S25N (Rab11S25N-GFP) proteins. Transfectant cells were incubated at 41°C for 14 h to accumulate VSVG in the ER. Then cells were incubated at 20°C for 2 h to allow the accumulation of VSVG in the TGN (VSVG, 20°C). VSVG transport to the cell surface was initiated by changing the temperature to a permissive condition (VSVG, 37°C). The expression of Rab11 versions was examined by the intrinsic fluorescence of GFP (GFP) fused to Rab11 proteins, and whereas VSVG expression was examined by the intrinsic fluorescence of YFP fused to VSVG protein. B, cells were grown in the presence of P4 for 5 days, allowing the cells to mainly eliminate GD3 from plasma membrane. Rab proteins (Rab11wt-GFP and Rab11S25N-GFP) were expressed by transient DNA transfection 2 days before removing the inhibitor P4. After 2 h at 20°C in the absence of the inhibitor, GD3-synthesized de novo was mainly accumulated at the TGN (GD3, 20°C). Transfer to 37°C for 2 h caused the GD3 to become progressively concentrated at the plasma membrane both in Rab 11-transfected and non-transfected cells (GD3, 37°C). GD3 expression was analyzed by confocal microscopy immunofluorescence in acetone-fixed cells using a monoclonal antibody R24 as the primary antibody and Alexa 546-conjugated goat anti-mouse IgG as the secondary antibody. As in A, the expression of Rab11 versions was examined by the intrinsic fluorescence of GFP (GFP) fused to Rab11 proteins. Arrows in B (GFP and GD3 rows) indicate cells doubly expressing Rab11 proteins and GD3. C, CHO-K1 cells stably expressing GD3 synthase tagged at the C terminus with the epitope HA (Sial-T2) were transiently transfected to express Rab11wt-GFP or Rab11S25N-GFP. The expression of Rab11 versions was examined by the intrinsic fluorescence of GFP (GFP) fused to Rab11 proteins. The expression of Sial-T2-HA was visualized by using a monoclonal antibody anti-HA and a goat anti-mouse, Alexa 546-conjugated secondary antibody. Arrows in C indicate cells doubly expressing Rab11 proteins and Sial-T2. Images shown in A and B are single xy confocal sections. Images shown in C were acquired with a Zeiss inverted microscopy equipped with epifluorescence. Scale bar, 20 m.

GD3 Exits the Golgi Complex Using a Brefeldin A-insensitive
to plasma membrane was markedly inhibited by BFA (ϳ60%). To examine the effect of BFA on GD3 exit from TGN, we first accumulated GD3 in the TGN (Fig. 4, E and F) following the protocol described in Fig. 1. As already observed, transfer of the cells to 37°C in the absence of BFA caused the GD3 to become progressively concentrated at the plasma membrane (Fig. 4G,  ϪBFA). Interestingly, BFA had no significant effect both on GD3 exit from TGN and transport to the plasma membrane (Fig. 4H,  ϩBFA). As a control of BFA action, we analyzed in all conditions the effect of BFA on the redistribution of Sial-T2 (proximal Golgi resident) into the ER (26) (data not shown). Our results indicate that in CHO-K1 cells operates a constitutive transport from the TGN to the cell surface, which is drastically inhibited by BFA. Regarding the lacking effect of BFA on GD3 exiting the TGN and exocytic transport, a preliminary interpretation supports the notion that in these processes BFA-sensitive GEF proteins might not participate, raising the existence of an alternative route(s) for glycolipid transport from Golgi to plasma membrane.
BFA Blocks the Exit from Golgi Complex of a Glycosylphosphatidylinositol-anchored Protein-To evaluate whether BFAsensitive GEFs proteins could operate in the exit and/or transport of another luminal membrane-bounded lipid, the effect of BFA in the delivery from TGN to plasma membrane of a chimeric protein containing a GPI-anchored signal fused to YFP (GPI-YFP) was investigated. GD3-expressing CHO-K1 cells were grown in the presence of P4 for 5 days. GPI-YFP was expressed by transient DNA transfection 1 day before removing the inhibitor P4. GPI-YFP-transfected CHO-K1 cells were incubated at 37°C for 24 h. Then cells were incubated at 20°C for 2 h to allow the accumulation of GPI-YFP at the TGN in media containing cycloheximide to prevent further synthesis of GPI-YFP (Fig. 5A, 20°C). Under this condition GPI-YFP was found co-localizing to a significant extent with UDP-GalNAc:LacCer/ GM3/GD3 N-acetylgalactosaminyltransferase, a well characterized TGN marker in CHO-K1 cells (26, 32) (data not shown).
To test the effect of BFA in the exit of GPI-YFP from Golgi complex, cells that had been blocked at 20°C were shifted to 37°C in the absence (ϪBFA) or presence (ϩBFA) of BFA for 1 h (Fig. 5A, 37°C). As a control of BFA action, we analyzed the effect of BFA on the redistribution of Sial-T2 (proximal Golgi resident) into the ER (Fig. 5B). Results clearly demonstrated that, unlike GD3 trafficking to plasma membrane, the constitutive transport of GPI-YFP in GD3-expressing CHO-K1 cells is rapidly inhibited upon treatment with BFA, strongly suggesting in this process an active role of BFA-sensitive GEF proteins.
Triton X-100 Solubility of GD3 during Biosynthetic Pathway-One mechanism of organizing cargo into domains is that of clustering. Accumulating evidence suggests that signals such as N-and O-linked glycans and GPI anchors help sequester proteins into GEM in the TGN, which is recognized by the sorting machinery as a signal for Golgi complex exiting and apical delivery (36). We recently described that an important fraction of plasma membrane GD3 resides in GEM. However, we were unable to find evidence for GEM localization of neither newly synthesized GD3 nor GD3 synthase in the proximal Golgi (24). These results imply that nascent GD3 segregates from its synthesizing transferases and then enters GEM. This event could have taken place along the secretory pathway and/or at the cell surface or shortly after synthesis in the Golgi cisternae (e.g. TGN). To determine the time and place at which GD3 becomes Triton X-100-insoluble on its way to the plasma membrane, ceramide (precursor for GD3 synthesis) was metabolically labeled for 12 h with [ 3 H]palmitate in the presence of the inhibitor P4 to block synthesis of GD3. Then culture medium was removed; cells were washed extensively to remove P4 and incubated with complete medium for 2 h at 20°C to synthesize and accumulate GD3 in TGN, as shown in Fig. 1B. After a 20°C treatment, cells were incubated at 37°C to allow GD3 to exit from TGN, and the detergent solubility/insolubility of GD3 was analyzed at different times. Before warm-up at 37°C (Fig. 6, 0 min), radioactive GD3 was under the limit of detection probably for conversion to GT3. At 7,15,30,45, and 90 min after warming to 37°C (Fig. 6), GD3 was found mainly in the soluble fraction after detergent extraction. In experiments not shown, we observed that at 15 h after warming to 37°C a significant amount of GD3 was insoluble to Triton X-100, strongly suggesting that the partition of GD3 is a long term process that under our experimental conditions principally occurs at the cell surface, according to the kinetic of GD3 transport shown in Fig. 1 and Table I. On the other hand, ϳ50% of GT3 was present in the Triton X-100-insoluble fraction before warm-up at 37°C. The insoluble fraction decreases at 7, 15, 30, and 45 min at 37°C, after which it increases at 90 min. DISCUSSION The topological orientation of ganglioside synthesis on the luminal surface of the Golgi complex implicates a priori a vesicular transport process of these glycosphingolipids from Golgi complex to cell surface. Supporting this notion, it was concluded from experiments with a cell-free assay system that transport of glycosphingolipids and glycoproteins through the Golgi complex stacks occurs by a vesicular process that is biochemically and kinetically similar for both types of molecules (37). Furthermore, it has been indicated that transport vesicles containing ganglioside GM3 are transported to the plasma membrane of CHO cells at a rate consistent with the rate of bulk flow of glycoproteins (38). On the basis of these antecedents, we have investigated in this work the exocytic transport of ganglioside GD3 from the Golgi complex to the plasma membrane of CHO-K1 cells.
Rab11 is a Ras-related small GTPase that has been shown to regulate the delivery of proteins from the TGN to the plasma membrane either through endosomes or by a direct transit (13). Although the dominant negative form of Rab11 efficiently affected the Golgi exiting and delivery of VSVG to the plasma membrane of CHO-K1 cells, it did not affect the efficiency of GD3 transport to the cell surface. Although we cannot absolutely discard with these experiments the existence of a GD3 transport dependent on Rab11 proteins, the Rab protein-independent route(s) is able to move almost all GD3 molecules from the TGN to plasma membrane in CHO-K1 cells.
Adaptins and GGAs proteins have been identified in trans elements of the Golgi complex and appear to function in sorting cargo into nascent clathrin-coated vesicles. The recruitment of these proteins to Golgi membrane is sensitive to BFA (33) and is most likely regulated by the TGN-localized and BFA-inhibited BFA-inhibited GEFs (34). At concentrations at which it blocks VSVG transport from the Golgi complex to the plasma membrane in CHO-K1 cells, BFA had no appreciable effect on GD3 transport from TGN to the cell surface, suggesting that in these processes BFA-sensitive GEF proteins might not participate.
It was described in chick embryo retina cells that the proximal-synthesized GD3 uses a transport mechanism dependent on ADP-ribosylation factor (ARF, BFA-affected transport) (39). The reasons for the apparent discrepancy between this result and those described in this work may be due to the conditions of metabolic GD3 labeling and BFA treatment, which is essential in the interpretation of these results since both studies could be discriminating against different transport stages or, perhaps, to the different cell types used. Accumulating evidence suggests that the site of synthesis of GD3 in CHO-K1 cells is the proximal Golgi. Thus, most of the N-linked oligosaccharides from Sial-T2 (GD3 synthase) were found to be both endoglycosidase H-sensitive and neuraminidase-insensitive, and in cells treated with BFA, a major fraction of the enzyme redistributed to the ER, indicative of a proximal Golgi location (26). Under our experimental conditions and given the abovementioned information, it is assumed that GD3 is mostly synthesized in the proximal Golgi and then conveyed and accumulated in the TGN at 20°C. At the end of this process the effect of BFA on TGN exit and transport of GD3 to the cell surface of FIG. 6. Triton X-100 solubility of GD3 during biosynthetic pathway. Ceramide (precursor for GD3 synthesis) was metabolically labeled for 12 h with [ 3 H]palmitate in the presence of the inhibitor P4 to block synthesis of GD3. Then culture medium was removed; cells were washed extensively to remove P4 and incubated with complete medium for 2 h at 20°C to synthesize and accumulate GD3 in TGN. After a 20°C treatment, cells were incubated at 37°C to allow GD3 exit from TGN. Membranes were extracted with 1% Triton X-100 at 4°C at different times (0, 7, 15, 30, 45, and 90 min), and lysates were centrifuged for 1 h to 100,000 ϫ g at 4°C. The supernatants (S, soluble fraction) were removed from the pellets (I, insoluble fraction), and gangliosides were purified and analyzed by HPTLC as indicated in "Materials and Methods." St, radioactive glycolipids obtained by metabolic labeling of GD3-expressing clone. The positions of glycolipids are indicated. The radioactive band labeled with an asterisk corresponds to a non-characterized lipid.
CHO-K1 cells was analyzed. On the contrary, in experiments done in chick embryo retina cells (39), BFA was present during the metabolic labeling of the proximal-synthesized GD3, raising the possibility that GD3 transport from the proximal Golgi to TGN is an ARF-mediated process in neural cells, whereas the exit of GD3 from the TGN remains to be evaluated in these cells. The same assumption could explain the results obtained by Warnock et al. (40). The authors described GM3 transport in CHO cells as inhibited by BFA. However, like experiments carried in neural cells (39), BFA was present during the metabolic labeling of the proximal-synthesized GM3.
It was hypothesized early that genesis of GEM occurs by a dynamic process in the TGN (6,41). Relying on the localization of GEM at the TGN and the affinity of glycosylphosphatidylinositol-anchored proteins for GEM (23)(24)(25), it was demonstrated that the GPI-linked proteins segregate into GEM-enriched carriers in the TGN and then are directed to the plasma membrane (25,42,43). More recently, an alternative model proposes that glycosylphosphatidylinositol-anchored protein exits the TGN in membrane-bound carriers that also contain non-resident GEM, although the two cargoes were laterally segregated (36). In this work we demonstrated that the constitutive transport of GPI-YFP in GD3-expressing CHO-K1 cells is rapidly inhibited upon treatment with BFA, strongly suggesting in this process an active role of BFA-sensitive GEF proteins.
It was described that high concentrations (Ͼ100 M) of D,Lthreo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol, a similar compound to P4, antagonizes the effect of BFA on retrograde membrane transport from Golgi to ER (44 -46) and inhibits anterograde membrane transport from the Golgi complex to the plasma membrane (47,48). It was suggested that the effect of D,L-threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol on these processes is probably attributed to an elevation of the intracellular ceramide and a reduction of sphingomyelin levels (47,48), changes in calcium homeostasis (44), or inhibition of BFA-induced ADP-ribosylation of specific proteins (45). In this work P4 depletes cellular gangliosides at a very low concentration (2.4 M) without causing both an increase in ceramide and a decrease in sphingomyelin levels. Moreover, P4 was removed before analyzing the effect of BFA, which effectively inhibited the constitutive transport of GPI-YFP from the Golgi to the plasma membrane and led to a dispersion of most proximal Golgi resident Sial-T2 throughout the ER. Taking these results into consideration, it is highly improbable that P4 antagonizes the effect of BFA on the GD3 exit from TGN and transport to the plasma membrane of CHO-K1 cells.
Like glycosylphosphatidylinositol-anchored proteins, GD3 was found to be a constituent of GEM or DRMs in the plasma membrane of CHO-K1 cell, but GD3 was almost absent from GEM when it was newly synthesized in the proximal Golgi (24). We took into consideration the possibility that GD3 could partition into GEM shortly after the synthesis in the Golgi cisternae (e.g. TGN), which could represent a sorting signal. Results shown in this work indicate that GD3 exits the TGN mainly excluded from DRMs and strongly suggests that partitioning of GD3 into DRMs at the TGN is not absolutely necessary for its exiting from the organelle. Moreover, it seems that alternative sorting signals could operate in the exiting of the glycolipid from the Golgi complex. Although we produced in CHO-K1 cells a block in glycosphingolipid synthesis, it does not seem to influence the distribution of membrane components on the cell surface of CHO-K1 cells, as already reported (23,49). Indeed, it was described that glycosphingolipids are not essential for the formation of DRMs and for efficient packaging and cell surface transport of placental alkaline phosphatase, a GPI-anchored protein resident in detergent-insoluble domains (50).
Concerning possible mechanisms of GD3 transport from Golgi to plasma membrane, we cannot entirely exclude the possibility that GD3 is transported to the plasma membrane by conventional vesicle coats. However, a plausible alternative is that the GD3 transport occurs by mechanisms other than coatmediated budding (8). In this sense the existence of a mechanism of constitutive export from the Golgi complex by formation of Golgi-to-plasma membrane carriers (GPCs) was recently suggested (51). GPCs emerge from the Golgi through a bulk flow process involving the formation of tubular TGN cargo domains that later detach en bloc from the TGN. The formation of GPCs does not involve cargo concentration and is not associated with the presence of known adaptors (AP1, AP2, AP4, and GGA) and coats (clathrin and ␤-COP) proteins on the GPC precursor. In light of these antecedents it is particularly attractive to investigate if GD3 conveys to plasma membrane by using GPC carriers and whether this putative mechanism of intracellular transport of GD3 is also used for other glycolipids. Finally, whatever the mechanism of GD3 exiting from Golgi complex, it is clear that lateral membrane segregation of VSVG (non-GEM resident) and GPI-anchored proteins (GEM resident) from GD3 is at least required before exiting TGN.